The application of decision theory to screening of driving under the influence (DUI) offenders is illustrated in an evaluation study that investigated the validity of a structured interview and a survey instrument. These findings are examined graphically using the relative operating characteristic (ROC). This graph relates the proportion of true-positive cases to the proportion of false-positive cases for various placements of the decision cut-off score. The ROC'S two principal parameters, test sensitivity and examiner bias, are used to provide a complete quantitative description of the performance of two screening instruments. A major goal of this study is to show how a new measure of examiner bias, the “cost ratio,” improves the evaluation of DUI screening programs. (Author/publisher)
Abstract